CUDA support improvements (#1168)

* Add test result for compile-to-cuda

* Add RAII for some CUDA types to simplify usage.

* First pass handling of some instrinsics on CUDA (for example transcendentals)

* CUDA working with built in intrinsics.

* Add missing CUDA prelude intrinsics.

* CUDA matches CPU output on simple-cross-compile.slang

* First pass at hlsl-scalar-float-intrinsic.slang test.

* Fix smoothstep impl on CUDA and CPU.

* Fixed step intrinsic on CUDA/CPU.

* Added operator[] to Matrix for C++, to allow row access.
Needs a fix for CUDA.

* Fixed warning on clang build.

3 files changed, 93 insertions, 0 deletions

diff --git a/tests/compute/hlsl-scalar-float-intrinsic.slang b/tests/compute/hlsl-scalar-float-intrinsic.slang
new file mode 100644
index 000000000..213db4b23
--- /dev/null
+++ b/tests/compute/hlsl-scalar-float-intrinsic.slang
@@ -0,0 +1,88 @@
+//TEST(compute):COMPARE_COMPUTE_EX:-cpu -compute 
+//TEST(compute):COMPARE_COMPUTE_EX:-slang -compute
+//TEST(compute):COMPARE_COMPUTE_EX:-slang -compute -dx12
+//TEST(compute, vulkan):COMPARE_COMPUTE_EX:-vk -compute
+//TEST(compute, vulkan):COMPARE_COMPUTE_EX:-cuda -compute
+
+//TEST_INPUT:ubuffer(data=[0 0 0 0], stride=4):out,name outputBuffer
+RWStructuredBuffer<int> outputBuffer;
+
+[numthreads(4, 1, 1)]
+void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
+{
+    int idx = int(dispatchThreadID.x);
+    
+    float f = idx * (1.0f / (4.0f - 1));
+
+    int it = 0;
+    float ft = 0.0f;
+    
+    // fmod
+    // ft += f % 0.5f;
+    
+    ft += sin(f);
+    ft += cos(f);
+    ft += tan(f);
+    
+    ft += asin(f);
+    ft += acos(f);
+    ft += atan(f);
+    
+    ft += atan2(f, 2.0);
+    
+    {
+        float sf, cf;
+        sincos(f, sf, cf);
+        
+        ft += sf;
+        ft += cf;
+    }
+    
+    ft += rcp(1.0 + f);
+    ft += sign(f - 0.5);
+    
+    ft += saturate(f * 4 - 2.0);
+    
+    ft += sqrt(f);
+    ft += rsqrt(1.0f + f);
+    
+    ft += exp2(f);
+    ft += exp(f);
+        
+        
+    ft += frac(f * 3);
+//    ft += ceil(f * 5 - 3);
+    
+    ft += floor(f * 10 - 7);
+    ft += trunc(f * 7);
+     
+   
+    ft += log(f + 10.0);
+    ft += log2(f * 3 + 2);
+    
+    // ft += log10(f * 10 + 4);
+    
+    ft += abs(f * 4 - 2.0f);
+    
+    ft += min(0.5, f);
+    ft += max(f, 0.75);
+
+    ft += pow(0.5, f);
+
+    ft += smoothstep(0.2, 0.7, f);
+    ft += lerp(-100, 100, f);
+
+
+    ft += clamp(f, 0.1, 0.3);
+
+    ft += step(f, 0.5);
+
+    int vi = asint(f - f) + idx;
+    
+    ft += float(vi);
+    
+    uint vu = asuint(f);
+    ft += asfloat(vu);
+    
+    outputBuffer[idx] = int(ft * 16);
+}
\ No newline at end of file
diff --git a/tests/compute/hlsl-scalar-float-intrinsic.slang.expected.txt b/tests/compute/hlsl-scalar-float-intrinsic.slang.expected.txt
new file mode 100644
index 000000000..04d17659c
--- /dev/null
+++ b/tests/compute/hlsl-scalar-float-intrinsic.slang.expected.txt
@@ -0,0 +1,4 @@
+FFFFFA3C
+FFFFFEEC
+3CA
+8C1
diff --git a/tests/compute/transcendental.slang b/tests/compute/transcendental.slang
index aa40da752..bde43ee38 100644
--- a/tests/compute/transcendental.slang
+++ b/tests/compute/transcendental.slang
@@ -1,3 +1,4 @@
+//TEST(compute):COMPARE_COMPUTE:-cuda
 //TEST(compute):COMPARE_COMPUTE:-cpu
 //TEST(compute):COMPARE_COMPUTE:
 //TEST(compute,vulcan):COMPARE_COMPUTE:-vk